Transducers or sensors play a very important role in industry measurement and control system. One of the key specifications of sensor is its degree of accuracy. The low precise sensor, which acquires a higher degree of accuracy, can be enhanced its cost performance. In this paper, a new neural networks-based stress-sensor accuracy enhancement scheme is proposed. The neural networks can be considered as nonlinear filter, which reduces the noise of low accuracy sensor. The neural network model with six input nodes, five hidden nodes, and one out nodes is chosen in sensor data filter. We provided with both the noisy signal from low accuracy sensor and noiseless signal from signal from high precise sensor. The neural networks can be trained to reduce the nosie level of senor as s signal noise filter. The practice uses show that our method can provide sensor greater accuracy and environment suitability.
|